The cerebellar microcircuit as an adaptive filter: experimental and computational evidence

Distribution of neural plasticity in cerebellum-dependent motor learning

The cerebellum is essential for some forms of motor learning. Two examples that provide useful experimental models are modification of the vestibulo-ocular reflex and classical conditioning of the nictitating membrane response (NMR) in the rabbit. There has been considerable analysis of these behavioral models and of conditioning of the eyelid blink reflex, which is similar in several respects to NMR conditioning but with some key differences in its control circuitry. The evidence is consistent with the suggestion that storage of these motor memories is to be found within the cerebellum and its associated brainstem circuitry. The cerebellum presents many advantages as a model system to characterize the cellular and molecular mechanisms underpinning behavioral learning. And yet, localizing the essential synaptic changes has proven to be difficult. A major problem has been to establish to what extent these neural changes are distributed through the cerebellar cortex, cerebellar nuclei, and associated brainstem nuclei. Inspired by recent theoretical work, here we review evidence that the distribution of plasticity across cortical and cerebellar nuclear (or brainstem vestibular system) levels for different learning tasks may be different and distinct. Our primary focus is on classical conditioning of the NMR and eyelid blink, and we offer comparisons with mechanisms for modifications of the vestibulo-ocular reflex. We describe a view of cerebellar learning that satisfies theoretical and empirical analysis.

The organization of plasticity in the cerebellar cortex: from synapses to control

The cerebellum is thought to play a critical role in procedural learning, but the relationship between this function and the underlying cellular and synaptic mechanisms remains largely speculative. At present, at least nine forms of long-term synaptic and nonsynaptic plasticity (some of which are bidirectional) have been reported in the cerebellar cortex and deep cerebellar nuclei. These include long-term potentiation (LTP) and long-term depression at the mossy fiber-granule cell synapse, at the synapses formed by parallel fibers, climbing fibers, and molecular layer interneurons on Purkinje cells, and at the synapses formed by mossy fibers and Purkinje cells on deep cerebellar nuclear cells, as well as LTP of intrinsic excitability in granule cells, Purkinje cells, and deep cerebellar nuclear cells. It is suggested that the complex properties of cerebellar learning would emerge from the distribution of plasticity in the network and from its dynamic remodeling during the different phases of learning. Intrinsic and extrinsic factors may hold the key to explain how the different forms of plasticity cooperate to select specific transmission channels and to regulate the signal-to-noise ratio through the cerebellar cortex. These factors include regulation of neuronal excitation by local inhibitory networks, engagement of specific molecular mechanisms by spike bursts and theta-frequency oscillations, and gating by external neuromodulators. Therefore, a new and more complex view of cerebellar plasticity is emerging with respect to that predicted by the original "Motor Learning Theory," opening issues that will require experimental and computational testing.

Questioning the cerebellar doctrine

The basic principles of cerebellar function were originally described by Flourens, Cajal, and Marr/Albus/Ito, and they constitute the pillars of what can be considered to be the classic cerebellar doctrine. In their concepts, the main cerebellar function is to control motor behavior, Purkinje cells are the only cortical neuron receiving and integrating inputs from climbing fiber and mossy-parallel fiber pathways, and plastic modification at the parallel fiber synapses onto Purkinje cells constitutes the substrate of motor learning. Yet, because of recent technical advances and new angles of investigation, all pillars of the cerebellar doctrine now face regular re-examination. In this review, after summarizing the classic concepts and recent disputes, we attempt to synthesize an integrated view and propose a revisited version of the cerebellar doctrine.

Decorrelation learning in the cerebellum: computational analysis and experimental questions

Many cerebellar models use a form of synaptic plasticity that implements decorrelation learning. Parallel fibers carrying signals positively correlated with climbing-fiber input have their synapses weakened (long-term depression), whereas those carrying signals negatively correlated with climbing input have their synapses strengthened (long-term potentiation). Learning therefore ceases when all parallel-fiber signals have been decorrelated from climbing-fiber input. This is a computationally powerful rule for supervised learning and can be cast in a spike-timing dependent plasticity form for comparison with experimental evidence. Decorrelation learning is particularly well suited to sensory prediction, for example, in the reafference problem where external sensory signals are interfered with by reafferent signals from the organism's own movements, and the required circuit appears similar to the one found to mediate classical eye blink conditioning. However, for certain stimuli, avoidance is a much better option than simple prediction, and decorrelation learning can also be used to acquire appropriate avoidance movements. One example of a stimulus to be avoided is retinal slip that degrades visual processing, and decorrelation learning appears to play a role in the vestibulo-ocular reflex that stabilizes gaze in the face of unpredicted head movements. Decorrelation learning is thus suitable for both sensory prediction and motor control. It may also be well suited for generic spatial and temporal coordination, because of its ability to remove the unwanted side effects of movement. Finally, because it can be used with any kind of time-varying signal, the cerebellum could play a role in cognitive processing.

Feedback control of learning by the cerebello-olivary pathway

The ability to anticipate future events and to modify erroneous anticipatory actions is crucial for the survival of any organism. Both theoretical and empirical lines of evidence implicate the cerebellum in this ability. It is often suggested that the cerebellum acquires "expectations" or "internal models." However, except in a metaphorical sense, the cerebellum, which consists of a set of interconnected nerve cells, cannot contain "internal models" or "have expectations." In this chapter, we try to untangle these metaphors by translating them back into neurophysiological cause and effect relationships. We approach this task from within the paradigm of classical conditioning, in which a subject, through repeated presentations of a conditional stimulus, followed by an unconditional stimulus, acquires a conditioned response. Importantly, the conditioned response is timed so that it anticipates the unconditioned response. Available neurophysiological evidence suggests that Purkinje cells, in the cerebellar cortex, generate the conditioned response. In addition, Purkinje cells provide negative feedback to the inferior olive, which is a relay for the unconditional stimulus, via the nucleo-olivary pathway. Purkinje cells can therefore regulate the intensity of the signal derived from the unconditional stimulus, which, in turn, decides subsequent plasticity. Hence, as learning progresses, the olivary signal will become weaker and weaker due to increasing negative feedback from Purkinje cells. Thus, in an important sense, learning-induced changes in Purkinje cell activity constitute an "expectation" or "anticipation" of a future event (the unconditional stimulus), and, consistent with theoretical models, future learning depends on the accuracy of this expectation.

Long-term potentiation of inhibitory synaptic transmission onto cerebellar Purkinje neurons contributes to adaptation of vestibulo-ocular reflex

Synaptic plasticity in the cerebellum is thought to contribute to motor learning. In particular, long-term depression (LTD) at parallel fiber (PF) to Purkinje neuron (PN) excitatory synapses has attracted much attention of neuroscientists as a primary cellular mechanism for motor learning. In contrast, roles of plasticity at cerebellar inhibitory synapses in vivo remain unknown. Here, we have investigated the roles of long-lasting enhancement of transmission at GABAergic synapses on a PN that is known as rebound potentiation (RP). Previous studies demonstrated that binding of GABAA receptor with GABAA receptor-associated protein (GABARAP) is required for RP, and that a peptide that blocks this binding suppresses RP induction. To address the functional roles of RP, we generated transgenic mice that express this peptide fused to a fluorescent protein selectively in PNs using the PN-specific L7 promoter. These mice failed to show RP, although they showed no changes in the basal amplitude or frequency of miniature IPSCs. The transgenic mice also showed no abnormality in gross cerebellar morphology, LTD, or other excitatory synaptic properties, or intrinsic excitability of PNs. Next, we attempted to evaluate their motor control and learning ability by examining reflex eye movements. The basal dynamic properties of the vestibulo-ocular reflex and optokinetic response, and adaptation of the latter, were normal in the transgenic mice. In contrast, the transgenic mice showed defects in the adaptation of vestibulo-ocular reflex, a model paradigm of cerebellum-dependent motor learning. These results together suggest that RP contributes to a certain type of motor learning.

Regulation of interneuron excitability by gap junction coupling with principal cells

Electrical coupling of inhibitory interneurons can synchronize activity across multiple neurons, thereby enhancing the reliability of inhibition onto principal cell targets. It is unclear whether downstream activity in principal cells controls the excitability of such inhibitory networks. Using paired patch-clamp recordings, we show that excitatory projection neurons (fusiform cells) and inhibitory stellate interneurons of the dorsal cochlear nucleus form an electrically coupled network through gap junctions containing connexin36 (Cxc36, also called Gjd2). Remarkably, stellate cells were more strongly coupled to fusiform cells than to other stellate cells. This heterologous coupling was functionally asymmetric, biasing electrical transmission from the principal cell to the interneuron. Optogenetically activated populations of fusiform cells reliably enhanced interneuron excitability and generated GABAergic inhibition onto the postsynaptic targets of stellate cells, whereas deep afterhyperpolarizations following fusiform cell spike trains potently inhibited stellate cells over several hundred milliseconds. Thus, the excitability of an interneuron network is bidirectionally controlled by distinct epochs of activity in principal cells.

HCN1 channels in cerebellar Purkinje cells promote late stages of learning and constrain synaptic inhibition

Neural computations rely on ion channels that modify neuronal responses to synaptic inputs. While single cell recordings suggest diverse and neurone type-specific computational functions for HCN1 channels, their behavioural roles in any single neurone type are not clear. Using a battery of behavioural assays, including analysis of motor learning in vestibulo-ocular reflex and rotarod tests, we find that deletion of HCN1 channels from cerebellar Purkinje cells selectively impairs late stages of motor learning. Because deletion of HCN1 modifies only a subset of behaviours involving Purkinje cells, we asked whether the channel also has functional specificity at a cellular level. We find that HCN1 channels in cerebellar Purkinje cells reduce the duration of inhibitory synaptic responses but, in the absence of membrane hyperpolarization, do not affect responses to excitatory inputs. Our results indicate that manipulation of subthreshold computation in a single neurone type causes specific modifications to behaviour.

Motor thalamus integration of cortical, cerebellar and basal ganglia information: implications for normal and parkinsonian conditions

Motor thalamus (Mthal) is implicated in the control of movement because it is strategically located between motor areas of the cerebral cortex and motor-related subcortical structures, such as the cerebellum and basal ganglia (BG). The role of BG and cerebellum in motor control has been extensively studied but how Mthal processes inputs from these two networks is unclear. Specifically, there is considerable debate about the role of BG inputs on Mthal activity. This review summarizes anatomical and physiological knowledge of the Mthal and its afferents and reviews current theories of Mthal function by discussing the impact of cortical, BG and cerebellar inputs on Mthal activity. One view is that Mthal activity in BG and cerebellar-receiving territories is primarily "driven" by glutamatergic inputs from the cortex or cerebellum, respectively, whereas BG inputs are modulatory and do not strongly determine Mthal activity. This theory is steeped in the assumption that the Mthal processes information in the same way as sensory thalamus, through interactions of modulatory inputs with a single driver input. Another view, from BG models, is that BG exert primary control on the BG-receiving Mthal so it effectively relays information from BG to cortex. We propose a new "super-integrator" theory where each Mthal territory processes multiple driver or driver-like inputs (cortex and BG, cortex and cerebellum), which are the result of considerable integrative processing. Thus, BG and cerebellar Mthal territories assimilate motivational and proprioceptive motor information previously integrated in cortico-BG and cortico-cerebellar networks, respectively, to develop sophisticated motor signals that are transmitted in parallel pathways to cortical areas for optimal generation of motor programmes. Finally, we briefly review the pathophysiological changes that occur in the BG in parkinsonism and generate testable hypotheses about how these may affect processing of inputs in the Mthal.

Distributed cerebellar plasticity implements adaptable gain control in a manipulation task: a closed-loop robotic simulation

Adaptable gain regulation is at the core of the forward controller operation performed by the cerebro-cerebellar loops and it allows the intensity of motor acts to be finely tuned in a predictive manner. In order to learn and store information about body-object dynamics and to generate an internal model of movement, the cerebellum is thought to employ long-term synaptic plasticity. LTD at the PF-PC synapse has classically been assumed to subserve this function (Marr, 1969). However, this plasticity alone cannot account for the broad dynamic ranges and time scales of cerebellar adaptation. We therefore tested the role of plasticity distributed over multiple synaptic sites (Hansel et al., 2001; Gao et al., 2012) by generating an analog cerebellar model embedded into a control loop connected to a robotic simulator. The robot used a three-joint arm and performed repetitive fast manipulations with different masses along an 8-shape trajectory. In accordance with biological evidence, the cerebellum model was endowed with both LTD and LTP at the PF-PC, MF-DCN and PC-DCN synapses. This resulted in a network scheme whose effectiveness was extended considerably compared to one including just PF-PC synaptic plasticity. Indeed, the system including distributed plasticity reliably self-adapted to manipulate different masses and to learn the arm-object dynamics over a time course that included fast learning and consolidation, along the lines of what has been observed in behavioral tests. In particular, PF-PC plasticity operated as a time correlator between the actual input state and the system error, while MF-DCN and PC-DCN plasticity played a key role in generating the gain controller. This model suggests that distributed synaptic plasticity allows generation of the complex learning properties of the cerebellum. The incorporation of further plasticity mechanisms and of spiking signal processing will allow this concept to be extended in a more realistic computational scenario.

Clusters of cerebellar Purkinje cells control their afferent climbing fiber discharge

Climbing fibers, the projections from the inferior olive to the cerebellar cortex, carry sensorimotor error and clock signals that trigger motor learning by controlling cerebellar Purkinje cell synaptic plasticity and discharge. Purkinje cells target the deep cerebellar nuclei, which are the output of the cerebellum and include an inhibitory GABAergic projection to the inferior olive. This pathway identifies a potential closed loop in the olivo-cortico-nuclear network. Therefore, sets of Purkinje cells may phasically control their own climbing fiber afferents. Here, using in vitro and in vivo recordings, we describe a genetically modified mouse model that allows the specific optogenetic control of Purkinje cell discharge. Tetrode recordings in the cerebellar nuclei demonstrate that focal stimulations of Purkinje cells strongly inhibit spatially restricted sets of cerebellar nuclear neurons. Strikingly, such stimulations trigger delayed climbing-fiber input signals in the stimulated Purkinje cells. Therefore, our results demonstrate that Purkinje cells phasically control the discharge of their own olivary afferents and thus might participate in the regulation of cerebellar motor learning.

Transient dynamics and rhythm coordination of inferior olive spatio-temporal patterns

The inferior olive (IO) is a neural network belonging to the olivo-cerebellar system whose neurons are coupled with electrical synapses and display subthreshold oscillations and spiking activity. The IO is frequently proposed as the generator of timing signals to the cerebellum. Electrophysiological and imaging recordings show that the IO network generates complex spatio-temporal patterns. The generation and modulation of coherent spiking activity in the IO is one key issue in cerebellar research. In this work, we build a large scale IO network model of electrically coupled conductance-based neurons to study the emerging spatio-temporal patterns of its transient neuronal activity. Our modeling reproduces and helps to understand important phenomena observed in IO in vitro and in vivo experiments, and draws new predictions regarding the computational properties of this network and the associated cerebellar circuits. The main factors studied governing the collective dynamics of the IO network were: the degree of electrical coupling, the extent of the electrotonic connections, the presence of stimuli or regions with different excitability levels and the modulatory effect of an inhibitory loop (IL). The spatio-temporal patterns were analyzed using a discrete wavelet transform to provide a quantitative characterization. Our results show that the electrotonic coupling produces quasi-synchronized subthreshold oscillations over a wide dynamical range. The synchronized oscillatory activity plays the role of a timer for a coordinated representation of spiking rhythms with different frequencies. The encoding and coexistence of several coordinated rhythms is related to the different clusterization and coherence of transient spatio-temporal patterns in the network, where the spiking activity is commensurate with the quasi-synchronized subthreshold oscillations. In the presence of stimuli, different rhythms are encoded in the spiking activity of the IO neurons that nevertheless remains constrained to a commensurate value of the subthreshold frequency. The stimuli induced spatio-temporal patterns can reverberate for long periods, which contributes to the computational properties of the IO. We also show that the presence of regions with different excitability levels creates sinks and sources of coordinated activity which shape the propagation of spike wave fronts. These results can be generalized beyond IO studies, as the control of wave pattern propagation is a highly relevant problem in the context of normal and pathological states in neural systems (e.g., related to tremor, migraine, epilepsy) where the study of the modulation of activity sinks and sources can have a potential large impact.

Ageing of internal models: from a continuous to an intermittent proprioceptive control of movement

To control the sensory-motor system, internal models mimic the transformations between motor commands and sensory signals. The present study proposed to assess the effects of physiological adult ageing on the proprioceptive control of movement and the related internal models. To this aim, one group of young adults and one group of older adults performed an ankle contralateral concurrent matching task in two speed conditions (self-selected and fast). Error, temporal and kinematic variables were used to assess the matching performance. The results demonstrated that older adults used a different mode of control as compared to the young adults and suggested that the internal models of proprioceptive control were altered with ageing. Behavioural expressions of these alterations were dependent upon the considered condition of speed. In the self-selected speed condition, this alteration was expressed through an increased number of corrective sub-movements in older adults as compared to their young peers. This strategy enabled them to reach a level of end-point performance comparable to the young adults' performance. In the fast speed condition, older adults were no more able to compensate for their impaired internal models through additional corrective sub-movements and therefore decreased their proprioceptive control performance. These results provided the basis for a model of proprioceptive control of movement integrating the internal models theory and the continuous and intermittent modes of control. This study also suggested that motor control was affected by the frailty syndrome, i.e. a decreased resistance to stressors, which characterises older adults.

Persistent posttetanic depression at cerebellar parallel fiber to Purkinje cell synapses

Plasticity at the cerebellar parallel fiber to Purkinje cell synapse may underlie information processing and motor learning. In vivo, parallel fibers appear to fire in short high frequency bursts likely to activate sparsely distributed synapses over the Purkinje cell dendritic tree. Here, we report that short parallel fiber tetanic stimulation evokes a ∼7–15% depression which develops over 2 min and lasts for at least 20 min. In contrast to the concomitantly evoked short-term endocannabinoid-mediated depression, this persistent posttetanic depression (PTD) does not exhibit a dependency on the spatial pattern of synapse activation and is not caused by any detectable change in presynaptic calcium signaling. This persistent PTD is however associated with increased paired-pulse facilitation and coefficient of variation of synaptic responses, suggesting that its expression is presynaptic. The chelation of postsynaptic calcium prevents its induction, suggesting that post- to presynaptic (retrograde) signaling is required. We rule out endocannabinoid signaling since the inhibition of type 1 cannabinoid receptors, monoacylglycerol lipase or vanilloid receptor 1, or incubation with anandamide had no detectable effect. The persistent PTD is maximal in pre-adolescent mice, abolished by adrenergic and dopaminergic receptors block, but unaffected by adrenergic and dopaminergic agonists. Our data unveils a novel form of plasticity at parallel fiber synapses: a persistent PTD induced by physiologically relevant input patterns, age-dependent, and strongly modulated by the monoaminergic system. We further provide evidence supporting that the plasticity mechanism involves retrograde signaling and presynaptic diacylglycerol.

An adaptive filter model of cerebellar zone C3 as a basis for safe limb control?

The review asks how the adaptive filter model of the cerebellum might be relevant to experimental work on zone C3, one of the most extensively studied regions of cerebellar cortex. As far as features of the cerebellar microcircuit are concerned, the model appears to fit very well with electrophysiological discoveries concerning the importance of molecular layer interneurons and their plasticity, the significance of long-term potentiation and the striking number of silent parallel fibre synapses. Regarding external connectivity and functionality, a key feature of the adaptive filter model is its use of the decorrelation algorithm, which renders it uniquely suited to problems of sensory noise cancellation. However, this capacity can be extended to the avoidance of sensory interference, by appropriate movements of, for example, the eyes in the vestibulo-ocular reflex. Avoidance becomes particularly important when painful signals are involved, and as the climbing fibre input to zone C3 is extremely responsive to nociceptive stimuli, it is proposed that one function of this zone is the avoidance of pain by, for example, adjusting movements of the body to avoid self-harm. This hypothesis appears consistent with evidence from humans and animals concerning the role of the intermediate cerebellum in classically conditioned withdrawal reflexes, but further experiments focusing on conditioned avoidance are required to test the hypothesis more stringently. The proposed architecture may also be useful for automatic self-adjusting damage avoidance in robots, an important consideration for next generation 'soft' robots designed to interact with people.

Beyond "all-or-nothing" climbing fibers: graded representation of teaching signals in Purkinje cells

Arguments about the function of the climbing fiber (CF) input to the cerebellar cortex have fueled a rabid debate that started over 40 years ago, and continues to polarize the field to this day. The origin of the controversy can be traced back to 1969, the year David Marr published part of his dissertation work in a paper entitled “A theory of cerebellar cortex.” In Marr’s theory, CFs play a key role during the process of motor learning, providing an instructive signal that serves as a “teacher” for the post-synaptic Purkinje cells. Although this influential idea has found its way into the mainstream, a number of objections have been raised. For example, several investigators have pointed out that the seemingly “all-or-nothing” activation of the CF input provides little information and is too ambiguous to serve as an effective instructive signal. Here, we take a fresh look at these arguments in light of new evidence about the peculiar physiology of CFs. Based on recent findings we propose that at the level of an individual Purkinje cell, a graded instructive signal can be effectively encoded via pre- or post-synaptic modulation of its one and only CF input.

Functional role of the cerebellum in gamma-band synchronization of the sensory and motor cortices

The cerebellum is an essential structure for the control of movement. It sends abundant ascending projections to the cerebral cortex via the thalamus, but its contribution to cortical activity remains largely unknown. Here we studied its influence on cortical neuronal activity in freely moving rats. We demonstrate an excitatory action of the cerebellum on the motor thalamus and the motor cortex. We also show that cerebellar inactivation disrupts the gamma-band coherence of local field potential between the sensory and motor cortices during whisking. In contrast, phase locking of neuronal activities to local gamma oscillations was preserved in the sensory and motor cortices by cerebellar inactivation. These results indicate that the cerebellum contributes to coordinated sensorimotor cortical activities during motor activation and thus participates in the multiregional cortical processing of information.

An integrator circuit in cerebellar cortex

The brain builds dynamic models of the body and the outside world to predict the consequences of actions and stimuli. A well-known example is the oculomotor integrator, which anticipates the position-dependent elasticity forces acting on the eye ball by mathematically integrating over time oculomotor velocity commands. Many models of neural integration have been proposed, based on feedback excitation, lateral inhibition or intrinsic neuronal nonlinearities. We report here that a computational model of the cerebellar cortex, a structure thought to implement dynamic models, reveals a hitherto unrecognized integrator circuit. In this model, comprising Purkinje cells, molecular layer interneurons and parallel fibres, Purkinje cells were able to generate responses lasting more than 10 s, to which both neuronal and network mechanisms contributed. Activation of the somatic fast sodium current by subthreshold voltage fluctuations was able to maintain pulse-evoked graded persistent activity, whereas lateral inhibition among Purkinje cells via recurrent axon collaterals further prolonged the responses to step and sine wave stimulation. The responses of Purkinje cells decayed with a time-constant whose value depended on their baseline spike rate, with integration vanishing at low (< 1 per s) and high rates (> 30 per s). The model predicts that the apparently fast circuit of the cerebellar cortex may control the timing of slow processes without having to rely on sensory feedback. Thus, the cerebellar cortex may contain an adaptive temporal integrator, with the sensitivity of integration to the baseline spike rate offering a potential mechanism of plasticity of the response time-constant.

Parallel fiber and climbing fiber responses in rat cerebellar cortical neurons in vivo

Over the last few years we have seen a rapidly increasing interest in the functions of the inhibitory interneurons of the cerebellar cortex. However, we still have very limited knowledge about their physiological properties in vivo. The present study provides the first description of their spontaneous firing properties and their responses to synaptic inputs under non-anesthetized conditions in the decerebrated rat in vivo. We describe the spike responses of molecular layer interneurons (MLI) in the hemispheric crus1/crus2 region and compare them with those of Purkinje cells (PCs) and Golgi cells (GCs), both with respect to spontaneous activity and responses evoked by direct electrical stimulation of parallel fibers (PFs) and climbing fibers (CFs). In agreement with previous findings in the cat, we found that the CF responses in the interneurons consisted of relatively long lasting excitatory modulations of the spike firing. In contrast, activation of PFs induced rapid but short-lasting excitatory spike responses in all types of neurons. We also explored PF input plasticity in the short-term (10 min) using combinations of PF and CF stimulation. With regard to in vivo recordings from cerebellar cortical neurons in the rat, the data presented here provide the first demonstration that PF input to PC can be potentiated using PF burst stimulation and they suggest that PF burst stimulation combined with CF input may lead to potentiation of PF inputs in MLIs. We conclude that the basic responsive properties of the cerebellar cortical neurons in the rat in vivo are similar to those observed in the cat and also that it is likely that similar mechanisms of PF input plasticity apply.

ADAPTIVE AND PREDICTIVE CONTROL OF A SIMULATED ROBOT ARM

In this work, a basic cerebellar neural layer and a machine learning engine are embedded in a recurrent loop which avoids dealing with the motor error or distal error problem. The presented approach learns the motor control based on available sensor error estimates (position, velocity, and acceleration) without explicitly knowing the motor errors. The paper focuses on how to decompose the input into different components in order to facilitate the learning process using an automatic incremental learning model (locally weighted projection regression (LWPR) algorithm). LWPR incrementally learns the forward model of the robot arm and provides the cerebellar module with optimal pre-processed signals. We present a recurrent adaptive control architecture in which an adaptive feedback (AF) controller guarantees a precise, compliant, and stable control during the manipulation of objects. Therefore, this approach efficiently integrates a bio-inspired module (cerebellar circuitry) with a machine learning component (LWPR). The cerebellar-LWPR synergy makes the robot adaptable to changing conditions. We evaluate how this scheme scales for robot-arms of a high number of degrees of freedom (DOFs) using a simulated model of a robot arm of the new generation of light weight robots (LWRs).

Multivoxel patterns reveal functionally differentiated networks underlying auditory feedback processing of speech

The everyday act of speaking involves the complex processes of speech motor control. An important component of control is monitoring, detection, and processing of errors when auditory feedback does not correspond to the intended motor gesture. Here we show, using fMRI and converging operations within a multivoxel pattern analysis framework, that this sensorimotor process is supported by functionally differentiated brain networks. During scanning, a real-time speech-tracking system was used to deliver two acoustically different types of distorted auditory feedback or unaltered feedback while human participants were vocalizing monosyllabic words, and to present the same auditory stimuli while participants were passively listening. Whole-brain analysis of neural-pattern similarity revealed three functional networks that were differentially sensitive to distorted auditory feedback during vocalization, compared with during passive listening. One network of regions appears to encode an "error signal" regardless of acoustic features of the error: this network, including right angular gyrus, right supplementary motor area, and bilateral cerebellum, yielded consistent neural patterns across acoustically different, distorted feedback types, only during articulation (not during passive listening). In contrast, a frontotemporal network appears sensitive to the speech features of auditory stimuli during passive listening; this preference for speech features was diminished when the same stimuli were presented as auditory concomitants of vocalization. A third network, showing a distinct functional pattern from the other two, appears to capture aspects of both neural response profiles. Together, our findings suggest that auditory feedback processing during speech motor control may rely on multiple, interactive, functionally differentiated neural systems.

New supervised learning theory applied to cerebellar modeling for suppression of variability of saccade end points

A new supervised learning theory is proposed for a hierarchical neural network with a single hidden layer of threshold units, which can approximate any continuous transformation, and applied to a cerebellar function to suppress the end-point variability of saccades. In motor systems, feedback control can reduce noise effects if the noise is added in a pathway from a motor center to a peripheral effector; however, it cannot reduce noise effects if the noise is generated in the motor center itself: a new control scheme is necessary for such noise. The cerebellar cortex is well known as a supervised learning system, and a novel theory of cerebellar cortical function developed in this study can explain the capability of the cerebellum to feedforwardly reduce noise effects, such as end-point variability of saccades. This theory assumes that a Golgi-granule cell system can encode the strength of a mossy fiber input as the state of neuronal activity of parallel fibers. By combining these parallel fiber signals with appropriate connection weights to produce a Purkinje cell output, an arbitrary continuous input-output relationship can be obtained. By incorporating such flexible computation and learning ability in a process of saccadic gain adaptation, a new control scheme in which the cerebellar cortex feedforwardly suppresses the end-point variability when it detects a variation in saccadic commands can be devised. Computer simulation confirmed the efficiency of such learning and showed a reduction in the variability of saccadic end points, similar to results obtained from experimental data.

Processing of multi-dimensional sensorimotor information in the spinal and cerebellar neuronal circuitry: a new hypothesis

Why are sensory signals and motor command signals combined in the neurons of origin of the spinocerebellar pathways and why are the granule cells that receive this input thresholded with respect to their spike output? In this paper, we synthesize a number of findings into a new hypothesis for how the spinocerebellar systems and the cerebellar cortex can interact to support coordination of our multi-segmented limbs and bodies. A central idea is that recombination of the signals available to the spinocerebellar neurons can be used to approximate a wide array of functions including the spatial and temporal dependencies between limb segments, i.e. information that is necessary in order to achieve coordination. We find that random recombination of sensory and motor signals is not a good strategy since, surprisingly, the number of granule cells severely limits the number of recombinations that can be represented within the cerebellum. Instead, we propose that the spinal circuitry provides useful recombinations, which can be described as linear projections through aspects of the multi-dimensional sensorimotor input space. Granule cells, potentially with the aid of differentiated thresholding from Golgi cells, enhance the utility of these projections by allowing the Purkinje cell to establish piecewise-linear approximations of non-linear functions. Our hypothesis provides a novel view on the function of the spinal circuitry and cerebellar granule layer, illustrating how the coordinating functions of the cerebellum can be crucially supported by the recombinations performed by the neurons of the spinocerebellar systems.

Cerebellar theta burst stimulation modulates short latency afferent inhibition in Alzheimer's disease patients

The dysfunction of cholinergic neurons is a typical hallmark in Alzheimer's disease (AD). Previous findings demonstrated that high density of cholinergic receptors is found in the thalamus and the cerebellum compared with the cerebral cortex and the hippocampus. We aimed at investigating whether activation of the cerebello-thalamo-cortical pathway by means of cerebellar theta burst stimulation (TBS) could modulate central cholinergic functions evaluated in vivo by using the neurophysiological determination of Short-Latency Afferent Inhibition (SLAI). We tested the SLAI circuit before and after administration of cerebellar continuous TBS (cTBS) in 12 AD patients and in 12 healthy age-matched control subjects (HS). We also investigated potential changes of intracortical circuits of the contralateral primary motor cortex (M1) by assessing short intracortical inhibition (SICI) and intracortical facilitation (ICF). SLAI was decreased in AD patients compared to HS. Cerebellar cTBS partially restored SLAI in AD patients at later inter-stimulus intervals (ISIs), but did not modify SLAI in HS. SICI and ICF did not differ in the two groups and were not modulated by cerebellar cTBS. These results demonstrate that cerebellar magnetic stimulation is likely to affect mechanisms of cortical cholinergic activity, suggesting that the cerebellum may have a direct influence on the cholinergic dysfunction in AD.

Integration of sensory quanta in cuneate nucleus neurons in vivo

Discriminative touch relies on afferent information carried to the central nervous system by action potentials (spikes) in ensembles of primary afferents bundled in peripheral nerves. These sensory quanta are first processed by the cuneate nucleus before the afferent information is transmitted to brain networks serving specific perceptual and sensorimotor functions. Here we report data on the integration of primary afferent synaptic inputs obtained with in vivo whole cell patch clamp recordings from the neurons of this nucleus. We find that the synaptic integration in individual cuneate neurons is dominated by 4-8 primary afferent inputs with large synaptic weights. In a simulation we show that the arrangement with a low number of primary afferent inputs can maximize transfer over the cuneate nucleus of information encoded in the spatiotemporal patterns of spikes generated when a human fingertip contact objects. Hence, the observed distributions of synaptic weights support high fidelity transfer of signals from ensembles of tactile afferents. Various anatomical estimates suggest that a cuneate neuron may receive hundreds of primary afferents rather than 4-8. Therefore, we discuss the possibility that adaptation of synaptic weight distribution, possibly involving silent synapses, may function to maximize information transfer in somatosensory pathways.

Gating of long-term depression by Ca2+/calmodulin-dependent protein kinase II through enhanced cGMP signalling in cerebellar Purkinje cells

Long-term depression (LTD) at parallel fibre synapses on a cerebellar Purkinje cell has been regarded as a cellular basis for motor learning. Although Ca(2+)/calmodulin-dependent protein kinase II (CaMKII) has been implicated in the LTD induction as an important Ca(2+)-sensing molecule, the underlying signalling mechanism remains unclear. Here, we attempted to explore the potential signalling pathway underlying the CaMKII involvement in LTD using a systems biology approach, combined with validation by electrophysiological and FRET imaging experiments on a rat cultured Purkinje cell. Model simulation predicted the following cascade as a candidate mechanism for the CaMKII contribution to LTD: CaMKII negatively regulates phosphodiesterase 1 (PDE1), subsequently facilitates the cGMP/protein kinase G (PKG) signalling pathway and down-regulates protein phosphatase 2A (PP-2A), thus supporting the LTD-inducing positive feedback loop consisting of mutual activation of protein kinase C (PKC) and mitogen-activated protein kinase (MAPK). This model suggestion was corroborated by whole-cell patch clamp recording experiments. In addition, FRET measurement of intracellular cGMP concentration revealed that CaMKII activation causes sustained increase of cGMP, supporting the signalling mechanism of LTD induction by CaMKII. Furthermore, we found that activation of the cGMP/PKG pathway by nitric oxide (NO) can support LTD induction without activation of CaMKII. Thus, this study clarified interaction between NO and Ca(2+)/CaMKII, two important factors required for LTD.

Hippocampal and cerebellar mossy fibre boutons - same name, different function

Over a century ago, the Spanish anatomist Ramón y Cajal described 'mossy fibres' in the hippocampus and the cerebellum, which contain several presynaptic boutons. Technical improvements in recent decades have allowed direct patch-clamp recordings from both hippocampal and cerebellar mossy fibre boutons (hMFBs and cMFBs, respectively), making them ideal models to study fundamental properties of synaptic transmission. hMFBs and cMFBs have similar size and shape, but each hMFB contacts one postsynaptic hippocampal CA3 pyramidal neuron, while each cMFB contacts ∼50 cerebellar granule cells. Furthermore, hMFBs and cMFBs differ in terms of their functional specialization. At hMFBs, a large number of release-ready vesicles and low release probability (<0.1) contribute to marked synaptic facilitation. At cMFBs, a small number of release-ready vesicles, high release probability (∼0.5) and rapid vesicle reloading result in moderate frequency-dependent synaptic depression. These presynaptic mechanisms, in combination with faster postsynaptic currents of cerebellar granule cells compared with hippocampal CA3 pyramidal neurons, enable much higher transmission frequencies at cMFB compared with hMFB synapses. Analysing the underling mechanisms of synaptic transmission and information processing represents a fascinating challenge and may reveal insights into the structure-function relationship of the human brain.

Adaptive filters and internal models: multilevel description of cerebellar function

Cerebellar function is increasingly discussed in terms of engineering schemes for motor control and signal processing that involve internal models. To address the relation between the cerebellum and internal models, we adopt the chip metaphor that has been used to represent the combination of a homogeneous cerebellar cortical microcircuit with individual microzones having unique external connections. This metaphor indicates that identifying the function of a particular cerebellar chip requires knowledge of both the general microcircuit algorithm and the chip's individual connections. Here we use a popular candidate algorithm as embodied in the adaptive filter, which learns to decorrelate its inputs from a reference ('teaching', 'error') signal. This algorithm is computationally powerful enough to be used in a very wide variety of engineering applications. However, the crucial issue is whether the external connectivity required by such applications can be implemented biologically. We argue that some applications appear to be in principle biologically implausible: these include the Smith predictor and Kalman filter (for state estimation), and the feedback-error-learning scheme for adaptive inverse control. However, even for plausible schemes, such as forward models for noise cancellation and novelty-detection, and the recurrent architecture for adaptive inverse control, there is unlikely to be a simple mapping between microzone function and internal model structure. This initial analysis suggests that cerebellar involvement in particular behaviours is therefore unlikely to have a neat classification into categories such as 'forward model'. It is more likely that cerebellar microzones learn a task-specific adaptive-filter operation which combines a number of signal-processing roles.

Differential effects of ketamine/xylazine anesthesia on the cerebral and cerebellar cortical activities in the rat

Cerebellum is a highly organized structure with a crystalline morphology that has always intrigued neuroscientists. Much of the cerebellar research has been conducted in anesthetized animals, particularly using ketamine/xylazine combination in rats. It is not clear how and to what extent the cerebellar cortical circuitry is affected by this anesthesia. In this study, we recorded spontaneous and evoked potentials from the cerebellar surface with chronically implanted, flexible-substrate, multielectrode arrays in rats and compared them to the signals simultaneously recorded from the motor cortex with similar electrodes. The power spectra and the intercontact coherence plots of the spontaneous activity in the awake-quiet animals extended up to 800 Hz in the cerebellum and only up to 200 Hz in the motor cortex. Ketamine/xylazine anesthesia suppressed most of the activity in the cerebellar cortex, which was in clear contrast to the motor cortex. In the awake cerebellum, large coherence values were observed between contact pairs as far apart as ∼2 mm. Otherwise, there was not a discernable relation between the coherence and the intercontact distance. These results suggest that the surface electrodes can provide much more detailed information about the state of neural circuits when they are used on the cerebellar cortex compared with the cerebral areas. This may be due to the proximity of the molecular layer cells to the pial surface in the cerebellum.

Cross-correlations between pairs of neurons in cerebellar cortex in vivo

In the present paper we apply a new neurophysiological technique to make single-electrode, dual loose-patch recordings from pairs of neuronal elements in the cerebellar cortex in vivo. The analyzed cell pairs consisted of an inhibitory molecular layer interneuron and a Purkinje cell (PC) or a Golgi cell and a granule cell, respectively. To detect the magnitude of the unitary inhibitory synaptic inputs we used histograms of the spike activity of the target cell, triggered by the spikes of the inhibitory cell. Using this analysis, we found that single interneurons had no detectable effect on PC firing, which could be explained by an expected very low synaptic weight of individual interneuron-PC connections. However, interneurons did have a weak delaying effect on the overall series of interspike intervals of PCs. Due to the very high number of inhibitory synapses on each PC, a concerted activation of the interneurons could still achieve potent PC inhibition as previously shown. In contrast, in the histograms of the Golgi cell-granule cell pairs, we found a weak inhibitory effect on the granule cell but only at the time period defined as the temporal domain of the slow IPSP previously described for this connection. Surprisingly, the average granule cell firing frequency sampled at one second was strongly modulated with a negative correlation to the overall firing level of the Golgi cell when the latter was modified through current injection via the patch pipette. These findings are compatible with that tonic inhibition is the dominant form of Golgi cell-granule cell inhibition in the adult cerebellum in vivo.

Modeling spike-train processing in the cerebellum granular layer and changes in plasticity reveal single neuron effects in neural ensembles

The cerebellum input stage has been known to perform combinatorial operations on input signals. In this paper, two types of mathematical models were used to reproduce the role of feed-forward inhibition and computation in the granular layer microcircuitry to investigate spike train processing. A simple spiking model and a biophysically-detailed model of the network were used to study signal recoding in the granular layer and to test observations like center-surround organization and time-window hypothesis in addition to effects of induced plasticity. Simulations suggest that simple neuron models may be used to abstract timing phenomenon in large networks, however detailed models were needed to reconstruct population coding via evoked local field potentials (LFP) and for simulating changes in synaptic plasticity. Our results also indicated that spatio-temporal code of the granular network is mainly controlled by the feed-forward inhibition from the Golgi cell synapses. Spike amplitude and total number of spikes were modulated by LTP and LTD. Reconstructing granular layer evoked-LFP suggests that granular layer propagates the nonlinearities of individual neurons. Simulations indicate that granular layer network operates a robust population code for a wide range of intervals, controlled by the Golgi cell inhibition and is regulated by the post-synaptic excitability.

Event time representation in cerebellar mossy fibres arising from the lateral reticular nucleus

Time representation is an important element of cerebellar neural processing, but the mechanisms involved are poorly understood. We demonstrate that the major mossy fibre input system originating from the lateral reticular nucleus (LRN) can represent sensory event timing over hundreds of milliseconds. In vivo, cerebellar-projecting LRN neurons discharge extremely regularly with a clock-like rhythm. In response to stimulation of a wide peripheral receptive field, firing briefly pauses then resumes with precise timing. The precision of post-stimulus spikes and the regularity of firing mean that the stimulus timing is represented by LRN spike timing over hundreds of milliseconds. In an arithmetic progression model of LRN neuron firing, highly predictable post-stimulus spike timing is modulated by changing the variability of the first post-inhibitory spike and of the subsequent interspike intervals. From in vitro analysis we show that the Ca(2+)-activated small-conductance K(+) current (SK) contributes to interspike interval regularity and that the hyperpolarization-activated cation current (I(h)) contributes to short-latency, high-precision post-hyperpolarisation spike timing. Consistent with this, we demonstrate in vivo that resumption of firing becomes more sharply timed after longer stimulus-evoked pauses. Thus, I(h) is a potential conductance that could mediate the precisely timed resumption of firing after the pause. Through the widespread projections of LRN neurons, these properties may enable the LRN to provide precisely timed signals to the cerebellum over a prolonged period following a stimulus, which may also both activate and sustain oscillatory processes in the cerebellar cortex.

Bio-inspired adaptive feedback error learning architecture for motor control

This study proposes an adaptive control architecture based on an accurate regression method called Locally Weighted Projection Regression (LWPR) and on a bio-inspired module, such as a cerebellar-like engine. This hybrid architecture takes full advantage of the machine learning module (LWPR kernel) to abstract an optimized representation of the sensorimotor space while the cerebellar component integrates this to generate corrective terms in the framework of a control task. Furthermore, we illustrate how the use of a simple adaptive error feedback term allows to use the proposed architecture even in the absence of an accurate analytic reference model. The presented approach achieves an accurate control with low gain corrective terms (for compliant control schemes). We evaluate the contribution of the different components of the proposed scheme comparing the obtained performance with alternative approaches. Then, we show that the presented architecture can be used for accurate manipulation of different objects when their physical properties are not directly known by the controller. We evaluate how the scheme scales for simulated plants of high Degrees of Freedom (7-DOFs).

An internal model architecture for novelty detection: implications for cerebellar and collicular roles in sensory processing

The cerebellum is thought to implement internal models for sensory prediction, but details of the underlying circuitry are currently obscure. We therefore investigated a specific example of internal-model based sensory prediction, namely detection of whisker contacts during whisking. Inputs from the vibrissae in rats can be affected by signals generated by whisker movement, a phenomenon also observable in whisking robots. Robot novelty-detection can be improved by adaptive noise-cancellation, in which an adaptive filter learns a forward model of the whisker plant that allows the sensory effects of whisking to be predicted and thus subtracted from the noisy sensory input. However, the forward model only uses information from an efference copy of the whisking commands. Here we show that the addition of sensory information from the whiskers allows the adaptive filter to learn a more complex internal model that performs more robustly than the forward model, particularly when the whisking-induced interference has a periodic structure. We then propose a neural equivalent of the circuitry required for adaptive novelty-detection in the robot, in which the role of the adaptive filter is carried out by the cerebellum, with the comparison of its output (an estimate of the self-induced interference) and the original vibrissal signal occurring in the superior colliculus, a structure noted for its central role in novelty detection. This proposal makes a specific prediction concerning the whisker-related functions of a region in cerebellar cortical zone A₂ that in rats receives climbing fibre input from the superior colliculus (via the inferior olive). This region has not been observed in non-whisking animals such as cats and primates, and its functional role in vibrissal processing has hitherto remained mysterious. Further investigation of this system may throw light on how cerebellar-based internal models could be used in broader sensory, motor and cognitive contexts.

Distributed synergistic plasticity and cerebellar learning

Studies on synaptic plasticity in the context of learning have been dominated by the view that a single, particular type of plasticity forms the underlying mechanism for a particular type of learning. However, emerging evidence shows that many forms of synaptic and intrinsic plasticity at different sites are induced conjunctively during procedural memory formation in the cerebellum. Here, we review the main forms of long-term plasticity in the cerebellar cortex that underlie motor learning. We propose that the different forms of plasticity in the granular layer and the molecular layer operate synergistically in a temporally and spatially distributed manner, so as to ultimately create optimal output for behaviour.

Cerebellar modulation of human associative plasticity

Paired associative stimulation (PAS) is a method commonly used in human studies of motor cortex synaptic plasticity. It involves repeated pairs of electrical stimuli to the median nerve and transcranial magnetic stimulation (TMS) of the motor cortex. If the interval between peripheral and TMS stimulation is around 21–25 ms, corticospinal excitability is increased for the following 30–60 min via a long term potentiation (LTP)-like effect within the primary motor cortex. Previous work has shown that PAS depends on the present and previous levels of activity in cortex, and that it can be modified by motor learning or attention. Here we show that simultaneous transcranial direct current stimulation (TDCS; 2 mA) over the cerebellum can abolish the PAS effect entirely. Surprisingly, the effect is seen when the PAS interval is 25 ms but not when it is 21.5 ms. There are two implications from this work. First, the cerebellum influences PAS effects in motor cortex; second, LTP-like effects of PAS have at least two different mechanisms. The results are relevant for interpretation of pathological changes that have been reported in response to PAS in people with movement disorders and to changes in healthy individuals following exercise or other interventions.

Neuroecology of cartilaginous fishes: the functional implications of brain scaling

It is a widely accepted view that neural development can reflect morphological adaptations and sensory specializations. The aim of this review is to give a broad overview of the current status of brain data available for cartilaginous fishes and examine how perspectives on allometric scaling of brain size across this group of fishes has changed within the last 50 years with the addition of new data and more rigorous statistical analyses. The current knowledge of neuroanatomy in cartilaginous fishes is reviewed and data on brain size (encephalization, n = 151) and interspecific variation in brain organization (n = 84) has been explored to ascertain scaling relationships across this clade. It is determined whether similar patterns of brain organization, termed cerebrotypes, exist in species that share certain lifestyle characteristics. Clear patterns of brain organization exist across cartilaginous fishes, irrespective of phylogenetic grouping and, although this study was not a functional analysis, it provides further evidence that chondrichthyan brain structures might have developed in conjunction with specific behaviours or enhanced cognitive capabilities. Larger brains, with well-developed telencephala and large, highly foliated cerebella are reported in species that occupy complex reef or oceanic habitats, potentially identifying a reef-associated cerebrotype. In contrast, benthic and benthopelagic demersal species comprise the group with the smallest brains, with a relatively reduced telencephalon and a smooth cerebellar corpus. There is also evidence herein of a bathyal cerebrotype; deep-sea benthopelagic sharks possess relatively small brains and show a clear relative hypertrophy of the medulla oblongata. Despite the patterns observed and documented, significant gaps in the literature have been highlighted. Brain mass data are only currently available on c. 16% of all chondrichthyan species, and only 8% of species have data available on their brain organization, with far less on subsections of major brain areas that receive distinct sensory input. The interspecific variability in brain organization further stresses the importance of performing functional studies on a greater range of species. Only an expansive data set, comprised of species that span a variety of habitats and taxonomic groups, with widely disparate behavioural repertoires, combined with further functional analyses, will help shed light on the extent to which chondrichthyan brains have evolved as a consequence of behaviour, habitat and lifestyle in addition to phylogeny.

Thin dendrites of cerebellar interneurons confer sublinear synaptic integration and a gradient of short-term plasticity

Interneurons are critical for neuronal circuit function, but how their dendritic morphologies and membrane properties influence information flow within neuronal circuits is largely unknown. We studied the spatiotemporal profile of synaptic integration and short-term plasticity in dendrites of mature cerebellar stellate cells by combining two-photon guided electrical stimulation, glutamate uncaging, electron microscopy, and modeling. Synaptic activation within thin (0.4 μm) dendrites produced somatic responses that became smaller and slower with increasing distance from the soma, sublinear subthreshold input-output relationships, and a somatodendritic gradient of short-term plasticity. Unlike most studies showing that neurons employ active dendritic mechanisms, we found that passive cable properties of thin dendrites determine the sublinear integration and plasticity gradient, which both result from large dendritic depolarizations that reduce synaptic driving force. These integrative properties allow stellate cells to act as spatiotemporal filters of synaptic input patterns, thereby biasing their output in favor of sparse presynaptic activity.

A computational mechanism for unified gain and timing control in the cerebellum

Precise gain and timing control is the goal of cerebellar motor learning. Because the basic neural circuitry of the cerebellum is homogeneous throughout the cerebellar cortex, a single computational mechanism may be used for simultaneous gain and timing control. Although many computational models of the cerebellum have been proposed for either gain or timing control, few models have aimed to unify them. In this paper, we hypothesize that gain and timing control can be unified by learning of the complete waveform of the desired movement profile instructed by climbing fiber signals. To justify our hypothesis, we adopted a large-scale spiking network model of the cerebellum, which was originally developed for cerebellar timing mechanisms to explain the experimental data of Pavlovian delay eyeblink conditioning, to the gain adaptation of optokinetic response (OKR) eye movements. By conducting large-scale computer simulations, we could reproduce some features of OKR adaptation, such as the learning-related change of simple spike firing of model Purkinje cells and vestibular nuclear neurons, simulated gain increase, and frequency-dependent gain increase. These results suggest that the cerebellum may use a single computational mechanism to control gain and timing simultaneously.